FINAL_Bench

Contact: arxivgpt@gmail.com

SeaWolf-AI updated a Space 19 minutes ago

SeaWolf-AI updated a Space 19 minutes ago

SeaWolf-AI new activity 24 minutes ago

Darwin Family: MRI-Trust-Weighted Evolutionary Merging for Training-Free Scaling of Language-Model Reasoning

published an article about 7 hours ago

about 7 hours ago

published an article 1 day ago

published an article 7 days ago

published an article 19 days ago

published an article 22 days ago

published an article 24 days ago

published an article 25 days ago

published an article about 1 month ago

published an article about 1 month ago

published an article about 1 month ago

published an article about 2 months ago

published an article 2 months ago

published an article 2 months ago

published an article 2 months ago

published an article 3 months ago

published an article 3 months ago

published an article 3 months ago

published an article 3 months ago

published an article 4 months ago

published an article 4 months ago

AI & ML interests

Recent Activity

Papers

Articles

The Sleeper in the Weights: We Planted a Backdoor in an Agent, and Only the Kernel Caught It

ViKernel: Why the Kill Switch Must Live Inside the Model

Pretraining a 325M Language Model from Zero on an Inference-Only NPU

Backpropagation on an Inference-Only NPU: Pretraining VKORE-46M-NPU From Scratch

The Open Quantum Challenge: Quantum Simulation and QEC Decoding on Classical GPUs

Leading the System One Mosaic Benchmark: What Darwin-27B-ZTC-v2's #1 Means

Darwin-27B-ZTC: A Single-Pass Judge and a Quantitative Look at Its Calibration

Data-center AI, now on a laptop — POCKET-Darwin-180B

Open Superconductor Challenge: Help Discover the Next 2D Superconductor — From Your Laptop

Your model already knows it's wrong. Asking costs 0.06 seconds and zero tokens.

Same bytes, closer to the original: two lines of AutoRound we had wrong

Did the Civilization Emerge, or Was It Recited?

Writing Down the Line Between Luck and Skill

We changed one line and the benchmark score moved 0.21 AUROC

Who Tells You Whether the Molecule Your AI Just Designed Is Any Good?

AX-Ray, Finding Causal-Leakage Defects in Two General-Purpose Public Models

The Fast Gemma Challenge: our verified-SOTA recipe, in full

POCKET: a 35-billion-parameter model that runs on your iPhone — and on your PC with no GPU

Aether-7B-5Attn: A 100% Open-Source Sovereign Foundation Model — and a Controlled Experiment in Heterogeneous Attention

VKUE: No GPU? Runs Anyway — a 34.7B Reasoner on a Laptop and on Bare CPU

Quantum Cryptanalysis on Real Hardware: Pushing Symmetric-Structure Key Recovery Beyond the Published Frontier

Adding a GPU Without Building One

Chitos: From Detection to Proof — An Autonomous Security AI That Actually Exploits

FINAL-Bench Quantum: An Open, Neutral Benchmark for Quantum-Computing Methods

Training-Free Reasoning at 88.89% on GPQA Diamond: How Darwin Family Hit Frontier Scores Without a Single Gradient Step

Darwin-TTS: We Gave a TTS Model 3% of an LLM's Brain — It Started Showing Emotion

"Darwin-27B-Opus: Surpassing the Foundation Model Without Training"

Darwin V6: Diagnostic-Guided Evolutionary Model Merging

"The Child That Surpassed Both Parents Through MRI-Guided Evolutionary Merge"

Introducing WM Bench: A Benchmark for Cognitive Intelligence in World Models

🏟️ Smol AI WorldCup: A 5-Axis Benchmark That Reveals What Small Language Models Can Really Do

MARL: Runtime Middleware That Reduces LLM Hallucination Without Fine-Tuning

Structural Problems in AI Benchmarking and the Case for a Unified Evaluation Framework

Do Bubbles Form When Tens of Thousands of AIs Simulate Capitalism?

FINAL Bench: The Real Bottleneck to AGI Is Self-Correction

Team members 1

FINAL-Bench's activity

The Sleeper in the Weights: We Planted a Backdoor in an Agent, and Only the Kernel Caught It

ViKernel: Why the Kill Switch Must Live Inside the Model

Pretraining a 325M Language Model from Zero on an Inference-Only NPU

Backpropagation on an Inference-Only NPU: Pretraining VKORE-46M-NPU From Scratch

The Open Quantum Challenge: Quantum Simulation and QEC Decoding on Classical GPUs

Leading the System One Mosaic Benchmark: What Darwin-27B-ZTC-v2's #1 Means

Darwin-27B-ZTC: A Single-Pass Judge and a Quantitative Look at Its Calibration

Data-center AI, now on a laptop — POCKET-Darwin-180B

Open Superconductor Challenge: Help Discover the Next 2D Superconductor — From Your Laptop

Your model already knows it's wrong. Asking costs 0.06 seconds and zero tokens.

Same bytes, closer to the original: two lines of AutoRound we had wrong

Did the Civilization Emerge, or Was It Recited?

Writing Down the Line Between Luck and Skill

We changed one line and the benchmark score moved 0.21 AUROC

Who Tells You Whether the Molecule Your AI Just Designed Is Any Good?

AX-Ray, Finding Causal-Leakage Defects in Two General-Purpose Public Models

The Fast Gemma Challenge: our verified-SOTA recipe, in full

POCKET: a 35-billion-parameter model that runs on your iPhone — and on your PC with no GPU

Aether-7B-5Attn: A 100% Open-Source Sovereign Foundation Model — and a Controlled Experiment in Heterogeneous Attention

VKUE: No GPU? Runs Anyway — a 34.7B Reasoner on a Laptop and on Bare CPU